Diffusing intelligence: building the app layer as AI becomes a utility

Summary

A strategic framework for companies building products on top of foundation models as AI capabilities become a commodity utility. The author argues that moats exist not in raw model capability but in systems that route and diffuse intelligence through specific industries, solving real problems end-to-end. The piece outlines seven tactical approaches to capture value: orchestrating multiplayer networks, accumulating proprietary workflow data, giving customers control over their own transformation, establishing domain-specific vision, continually climbing the abstraction layer, pricing against new outcomes rather than displaced labour, and becoming structurally necessary to the economy. Read the full thread on X.

Key Claims

• Adoption and economic transformation decouple in time — AI adoption may appear widespread but productivity gains lag institutional restructuring by years, creating a narrow arbitrage window for product builders.
• Product moats shift from defending against raw intelligence to coordination systems where humans and agents collaborate on workflows no single model can run alone.
• Value accrues not to labs with the best models but to companies that compound domain-specific data, customer history, and process corrections; the more they accumulate, the harder they are to replace.
• Product builders must move upmarket as commodity capabilities erode the bottom: shift from serving individual users to line managers to executives, or watch the layer underneath get eaten by model improvements and cheaper alternatives.
• Shift pricing from seat-based or productivity-per-human-replaced to outcome-based (tickets closed, revenue generated, drugs in trial), capturing the unbounded ROI of tasks bottlenecked on human attention rather than labour cost.
• Strategic necessity wins where labs cannot compete: companies that become trusted, neutral intermediaries between AI and regulated industries, or hold institutions accountable in ways a model API cannot, become economically essential.

Quotes

• "Moats are useless to defend against a flood. But dams & waterways that direct the water & power to crops, reservoirs, and homes that otherwise wouldn't have it are critical infrastructure."
• "The real world is tangled up in incentives, approvals, accountability, edge cases, legacy systems, and humans coordinating with other humans."
• "Cloud didn't stop Stripe, Uber, Doordash, Salesforce, Workday, ServiceNow, or Shopify from becoming generational businesses. Abundant intelligence is the mother of all platforms."